GPU networking using an integrated command processor

Active Publication Date: 2019-05-16
ADVANCED MICRO DEVICES INC
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Benefits of technology

The patent describes a system and method for generating network messages directly from a parallel processor, such as a CPU and a GPU, without involving a general purpose processor. This is achieved by using a data parallel processor with a highly parallel architecture and a command processor that supports remote direct memory access (RDMA) networking. The system includes a computing system with a processor, input / output interfaces, bus, memory device(s), and network interface(s). The technical effect of this invention is improved performance and efficiency in generating network messages and reducing delays in computations.

Problems solved by technology

A data parallel processor incurs delays in computations while waiting for the CPU to coordinate network communication on its behalf.
Using the host to generate network messages for the GPU can potentially involve several back-to-back round trips from the GPU to the host and from the host to the network interface, resulting in a reduction in performance.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • GPU networking using an integrated command processor
  • GPU networking using an integrated command processor
  • GPU networking using an integrated command processor

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0009]In the following description, numerous specific details are set forth to provide a thorough understanding of the methods and mechanisms presented herein. However, one having ordinary skill in the art should recognize that the various embodiments may be practiced without these specific details. In some instances, well-known structures, components, signals, computer program instructions, and techniques have not been shown in detail to avoid obscuring the approaches described herein. It will be appreciated that for simplicity and clarity of illustration, elements shown in the figures have not necessarily been drawn to scale. For example, the dimensions of some of the elements may be exaggerated relative to other elements.

[0010]Various systems, apparatuses, methods, and computer-readable mediums for generating network messages directly from a parallel processor are disclosed herein. In one embodiment, a system includes at least a first processor, a second processor, and a network ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

Systems, apparatuses, and methods for generating network messages on a parallel processor are disclosed. A system includes at least a parallel processor, a general purpose processor, and a network interface unit. The parallel processor includes at least a plurality of compute units, a command processor, and a cache. A thread within a kernel executing on a compute unit of the parallel processor generates a network message and stores the network message and a corresponding indication in the cache. In response to detecting the indication of the network message in the cache, the command processor processes and conveys the network message to the network interface unit without involving the general purpose processor.

Description

BACKGROUNDDescription of the Related Art[0001]A general-purpose processor, such as a central processing unit (CPU), typically performs input / output (I / O) operations for a software application. In a system that includes multiple processors, the CPU might be the only processor that can generate network messages. The system can also include a data parallel processor in addition to the general-purpose processor. The data parallel processor has a highly parallel execution architecture. Data parallel processors can include graphics processing units (GPUs), digital signal processors (DSPs), and so forth. A data parallel processor incurs delays in computations while waiting for the CPU to coordinate network communication on its behalf. In a computing system with a CPU and a GPU, the CPU is often referred to as “the host”. Current GPU networking applications either require the host to pre-register or initiate network messages on behalf of the GPU. Using the host to generate network messages ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F9/54G06F9/30H04L29/12
CPCG06F9/542G06F9/546G06F9/3009H04L61/10G06F2015/765
InventorLEBEANE, MICHAEL WAYNEHAMIDOUCHE, KHALEDBENTON, WALTER B.
OwnerADVANCED MICRO DEVICES INC